Not to derail too much but this is the reason people are so overdiagnosed these days. All the criteria are vague, even psychiatrists overdiagnose all the time.
It's pattern recognition, like the other diagnostic fields, but more qualitative due to the human behavior inputs. This leads to larger margins of error, and equating it with tarot reading isn't an apt analogy.
Even the guy who chaired the DSM-IV thinks so: http://www.amazon.com/Saving-Normal-Out-Control-Medicalizati...
Modern psychiatry is 50% science 50% tarot cards if you ask me.